The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from GitHub PR comments and CI logs, which is essential for its automation functionality.
Ingestion points: PR reviews and thread comments are retrieved in scripts/pr-state.sh, and CI failure logs are accessed in scripts/classify-failure.sh.
Boundary markers: The skill requires the agent to obtain explicit user confirmation before responding to human reviewers, serving as a significant safety boundary.
Capability inventory: The skill can perform potentially sensitive actions such as pushing commits, merging PRs, and resolving review threads.
Sanitization: Data is parsed using jq for structured handling, though raw text content is eventually reviewed by the agent for decision-making.
[COMMAND_EXECUTION]: The skill utilizes local bash scripts and the GitHub CLI to perform PR management tasks.
Evidence: Use of gh pr view, gh pr checks, gh api graphql, and git push is documented in the provided scripts and instructions as the primary method of interaction.

babysit-pr